“But not all scientific applications have been able to harness that power,” says Andrew Chien, a senior scientist in Argonne’s Mathematics and Computer Science Division, William Eckhardt Distinguished Service Professor at the University of Chicago, and director of the CERES Center for Unstoppable Computing.
In his new book Computer Architecture for Scientists: Principles and Performance, Chien discusses four key aspects of computer performance: size (scaling to drive performance), implicit parallelism (helping sequential programs), dynamic locality (rearranging data), and teamwork (increasing performance through collaborations).
The book begins with a look at how computing is transforming society. It then turns to the structure of the computer and why structure matters in achieving performance. Topics range from general-purpose computers to multicore computers and the role of accelerators in customizing architectures for performance.
By providing a high-level view of computer architecture and the design principles needed for better performance, the book gives software developers and data scientists insights that they can apply in their applications.
“It’s not enough to just do great research,” Chien said. “Making computing knowledge clear and accessible to broader scientific communities is a critical task for computer scientists.”
That Chien has succeeded in this task is clear from the reviews, which praise the new book’s clarity and timeliness.
Book details: Andrew A. Chien, Computer Architecture for Scientists: Principles and Performance, Cambridge University Press, ISBN: 978-1-31651853-3.